01What the ASO audit skill checks
The skill is a set of instructions, not a program. It tells Claude which checks make up an audit, which Storelift MCP tool answers each one, and what to write when no tool can. Every check is run per storefront — one country on one platform — because a rank in Türkiye says nothing about the US.
| Area | With the MCP server | Without it |
|---|---|---|
| Keyword coverage | get_keywords: ranked, measured but absent, and not measured, counted separately; iOS demand per term | Every word of each target term appears across name, subtitle and keyword field |
| Rivals | get_rivals: the apps above you, on which terms, and their rating counts | UNKNOWN |
| Rank moves | get_history lined up with the dated changes from get_store_page | UNKNOWN |
| Listing | get_store_page: ratings and histogram, last version and notes, screenshots per device shelf, in-app events, editorial placements, similar-apps shelf | Field lengths, repeated words, Apple product names, pricing language |
| Reviews | get_reviews: recurring complaints with counts, new reviews since the last measurement | Recent low ratings read from the public page |
| Charts | get_charts (App Store only) | UNKNOWN |
| AI visibility | get_ai_visibility, measured on the Studio plan | UNKNOWN |
| Localization | Storefronts the app is tracked in | Which storefronts also index a second localization, and whether it is filled |
The skill uses all eight tools of the server — list_apps first, then the ones above — and no others. It never calls a tool the server does not have.
02Installing the skill in Claude Code and the Claude apps
The skill is a single file: download SKILL.md. It starts with the YAML frontmatter every Agent Skill carries — a name (aso-audit) and a description that tells Claude when to use it — followed by the instructions. Read it before you install it; it is plain text.
Claude Code
Claude Code loads personal skills from ~/.claude/skills/<name>/SKILL.md, available in every project:
mkdir -p ~/.claude/skills/aso-audit
curl -fsSL https://storelift.net/data/aso-audit-skill/SKILL.md -o ~/.claude/skills/aso-audit/SKILL.mdTo share it with a team through the repository, put it under the project's .claude/skills/ instead:
mkdir -p .claude/skills/aso-audit
curl -fsSL https://storelift.net/data/aso-audit-skill/SKILL.md -o .claude/skills/aso-audit/SKILL.mdClaude picks the skill up when a request matches its description, such as “run an ASO audit of my app”. You can also call it by name with /aso-audit.
Claude desktop and web apps
The Claude apps take custom skills as a ZIP file of the skill folder. Put SKILL.md in a folder named aso-audit, zip the folder, and upload it from the Skills area of the app's settings. Menu names change between app versions; the file stays the same.
Connecting the measured mode
The measured mode needs the Storelift MCP server in the same client. In Claude Code, add the hosted endpoint and sign in with /mcp:
claude mcp add --transport http storelift https://storelift.net/mcpClaude Desktop setup, the npm package and the Claude.ai connector are on the MCP server guide. The server needs a Pro or Studio plan; the checklist mode needs no account.
03Example prompts
- “Run an ASO audit of my app in the US and Germany App Stores.”
- “Audit my Google Play listing in Brazil and tell me what to fix first.”
- “Which high-demand keywords am I absent for, and who ranks there instead?”
- “My rank dropped last week. Check the listing timeline and reviews for a cause.”
- “Here is my name, subtitle and keyword field. Check them against the App Store rules.”
- “Give me an ASO audit report I can send to my team.”
The first four use measured data and need the MCP server. The last two work in either mode; in checklist mode, paste the keyword field yourself, because Apple does not publish it and no tool can read it.
04What the audit report looks like
One report per storefront. It opens with the mode (measured, with the measurement time, or checklist) and a count of checks: how many passed, how many need fixing, how many could not be checked.
There is no single ASO score, on purpose. A weighted score hides which part failed, and any weights would be invented. Counts keep each check visible.
- Fix first — ordered by evidence: a high-demand term you are absent for comes before a wording change, and each item names the term, rank or rule it rests on.
- Keywords — ranked 1–10, 11–50, below 50, absent and not measured, as five separate counts, then the term table.
- Rivals above you, listing, reviews, charts and AI visibility.
- Not checked — every UNKNOWN, with what would answer it.
Without the MCP server, the skill points each UNKNOWN to a free tool that answers it one term at a time: rank check, keyword difficulty, the keyword field packer and the character counter.
05What the skill cannot do
- Read App Store Connect or Play Console data — impressions, conversion rate, downloads, revenue. The MCP server reads the public stores, not your developer account. See App Store MCP server: two kinds.
- Read any app's iOS keyword field, yours or a rival's. Apple does not publish it.
- Read Apple Ads data. The server has no Apple Ads tools.
- Report ranks for terms you do not track, or history from before a term was added.
- Change anything. The skill writes a report; the listing and the tracking setup stay as they are.
It also will not estimate what it cannot read. When a number is missing, the report says UNKNOWN and names the data that would fill it, because an estimate printed next to measured numbers looks exactly like one of them.
Frequently asked
Is the ASO audit skill free?
Yes. The SKILL.md file is free to download and use. The checklist mode needs nothing else. The measured mode uses Storelift's MCP server, which needs a Pro or Studio plan.
Does it work in Claude Code?
Yes. Save it as ~/.claude/skills/aso-audit/SKILL.md for all projects, or under the project's .claude/skills/. Claude uses it when you ask for an ASO audit, or you can run /aso-audit.
Does it give my app an ASO score?
No. It reports how many checks passed, need fixing or could not be checked, per storefront. A weighted score would need invented weights and would hide which check failed.
Can it audit a competitor's app?
Partly. In checklist mode it can review any public listing you point it to. The measured tools read the apps tracked in your Storelift account; rivals appear there as the apps ranking above you, with their ranks and rating counts.
Does it change my App Store listing?
No. The skill and the MCP server are read-only. The report is text; what you ship stays your decision.
See this for your own app
Paste your store link, add your keywords. Rank, the apps above you and demand are measured every day.
Free plan: 1 app, 25 keywords in each of 2 storefronts, 30 days of history, no card, no end date. Pro is $20/mo.
Start free →